Abstract
All-optical in-memory computing is emerging as a critical technology for next-generation energy-efficient and high-speed information processing because it avoids frequent optical-electrical-optical conversions and integrates sensing, processing, and memory within a single device. Here, we report the demonstration of bidirectional polarization switching in a van der Waals heterostructure composed of ferroelectric CuInP2S6 (CIPS) and semiconducting MoS2. Wavelength-tunable excitation (660–405 nm) enables robust, bidirectional polarization reversal through the interaction between the photogenerated charges in MoS2/CIPS heterostructure and ferroelectric polarization charges in CIPS. Two wavelength-dependent carrier dynamic mechanisms were established specifically for excitations below and above the CIPS bandgap. These mechanisms result in opposite charge accumulation at the interface, leading to opposite polarization switching directions. The device demonstrates high-performance all-optical nonvolatile memory. Furthermore, it emulates all-optical controlled retina-like synaptic plasticity, including paired-pulse facilitation/inhibition, short-term and long-term potentiation and depression, and learning-forgetting behaviours, with wavelength-selective long-term potentiation and depression enabling neuromorphic image recognition. Additionally, the single device implements reconfigurable all-optical controlled Boolean logic gates.
| Original language | English |
|---|---|
| Article number | e22092 |
| Journal | Advanced Science |
| Volume | 13 |
| Issue number | 20 |
| DOIs | |
| State | Published - 9 Apr 2026 |
UN SDGs
This output contributes to the following UN Sustainable Development Goals (SDGs)
-
SDG 7 Affordable and Clean Energy
Keywords
- all-optical bidirectional polarization switching
- artificial visual system
- logic-in-memory
- neuromorphic computing
- van der Waals heterostructure
Fingerprint
Dive into the research topics of 'All-Optical Control of Bidirectional Polarization Switching in Ferroelectric Heterostructures for Neuromorphic and In-Memory Computing'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver